JP3928992B2 - Magnetic resonance imaging system - Google Patents

Description

この静磁場分布計測法は、ピクセルの静磁場強度や静磁場均一度を元に、水以外のスペクトル位置を修正するのに適用されている。
【００１１】
【発明が解決しようとする課題】
しかし、３Ｄ／４Ｄ−ＣＳＩ法を生体シミングの予備撮像として適用する場合、計測に長時間を要するという欠点がある。例えば３次元体積の計測には、位相エンコードの反復ループが３重になるため、マトリクス数の３乗回の励起が必要になる。一例として反復時間（ＴＲ）を１秒とし、マトリクスを１６×１６×１６とすると１秒×１６×１６×１６＝約６８分が必要になる。
【００１２】
【課題を解決するための手段】
上記課題を解決するため、本発明では静磁場分布測定およびシム電流決定を含む生体シミングにおいて、静磁場分布測定方法として、３次元高速ＭＲスペクトロスコピックイメージング法（MR Spectroscopic Imaing：以下ＭＲＳＩという）法を採用する（P.Mansfieldによる論文「ＮＭＲにおける化学シフトの空間マッピング」（Spatial Mapping of the Chemical Shift in NMR）、マグネティック レゾナンス イン メディスン（Magn. Reson. Med.）1,370-386（1984）や松井らによる論文「高速高空間分解能ＮＭＲスペクトロスコピー」（High-Speed Spatially Resolved High-Resolution NMR Spectroscoy）、J.Am.Chem.Soc.,107,2817-2818（1985）を参照のこと）。また、シム電流を決定するに際し、静磁場分布の展開を２次項までとする。これにより、実用的で迅速な静磁場均一化方法を提供する。
【００１３】
即ち、本発明の磁気共鳴イメージング装置は、静磁場を発生する静磁場発生手段と、前記静磁場の不均一をその空間成分毎に補正するシミング手段と、前記静磁場空間のスライス方向、位相エンコード方向、及びリードアウト方向にそれぞれスライス傾斜磁場、位相エンコード傾斜磁場、及びリードアウト傾斜磁場の各傾斜磁場を発生する傾斜磁場発生手段と、前記静磁場空間に配置された被検体を組成する原子の原子核スピンに核磁気共鳴を誘起するための高周波磁場を発生する高周波磁場発生手段と、前記被検体からの核磁気共鳴により生ずるエコー信号を受信する受信手段と、前記受信された核磁気共鳴信号から前記被検体の画像を再構成する信号処理手段と、所定のパルスシーケンスに基づいて前記傾斜磁場発生手段と前記高周波磁場発生手段と前記受信手段と前記信号処理手段を制御する計測制御手段とを備え、
前記計測制御手段は、前記高周波磁場の照射により前記被検体の所望の体積部分を励起し、次いで所定の静磁場不均一検出用パルスシーケンスにより、前記体積部分についての静磁場不均一を表すエコー信号を取得し、
前記信号処理手段は、前記静磁場不均一を表すエコー信号から前記体積部分の静磁場分布を得、この静磁場分布を前記シミング手段が発生する空間成分毎の磁場分布で展開し、該静磁場分布を均一化するように前記シミング手段の空間成分毎に流す電流の最適値を求め、
前記シミング手段は、その空間成分毎に前記最適値の電流を流す磁気共鳴イメージング装置において、
前記静磁場不均一検出用パルスシーケンスは、前記スライス方向にスライスエンコード傾斜磁場を、位相エンコード方向に位相エンコード傾斜磁場を、リードアウト方向に連続反転する傾斜磁場をそれぞれ印加する基本シーケンスを繰り返し、
前記信号処理手段は、前記リードアウト方向傾斜磁場が同一極性であって時間軸方向において同一間隔となるように取得された前記静磁場不均一を表すエコー信号から前記被検体のボクセル毎のスペクトルを求め、その特定ピークの共鳴周波数を静磁場強度に換算することにより、前記ボクセル毎の静磁場分布を得る。
【００１４】
３次元ＭＲＳＩ法は、リードアウト方向の傾斜磁場を連続反転させ、空間情報とスペクトル情報を重畳させて取得することにより、反復ループを１次元分削減した高速スペクトロスコピックイメージング方法であり、短時間にマルチボクセルスペクトルを取得することができる。
【００１５】
特定のスペクトルピークとしては水プロトンの共鳴周波数を用いることが好ましい。これにより生体内の概ね全てのボクセルに亘り静磁場強度を計測できる。また短時間化のために、繰り返し時間（ＴＲ）を短縮した場合にも高周波磁場として低フリップ角のパルスを用いることにより、高信号が得られる。更に３次元ＭＲＳＩ法としてスピンエコー（ＳＥ）法を基本としたシーケンスを採用しＴＥを長く設定した場合に、Ｔ２の差を利用して束縛水等からの信号を消去したシャープなスペクトルを得ることができ、共鳴周波数の読み取り精度が向上する。
【００１６】
また、本発明の好適な態様においては、前記信号処理手段は、前記ボクセル毎の静磁場分布を球面調和関数の１次項および２次項を用いて展開して、それぞれの空間成分毎に前記最適電流値を求め、前記シミング手段は、概略球面調和関数の１次項および２次項を生成する少なくとも８つの空間成分を有して、その空間成分毎に対応する前記最適電流を流す。本発明においてシミング手段として、１次項の磁場分布を有する傾斜磁場コイル及び概略２次項以上の高次磁場分布を有するシムコイルを利用する。
【００１７】
２次以下の項のみを用いて静磁場分布を近似することにより、小数のボクセルの静磁場計測データで展開項を決定でき、静磁場計測を実際的な短い時間内に行うことができる。また２次項以下とすることにより、シミング手段として特定の高次項のみを発生するシムコイルを製作する困難性を回避でき、また一般に３次以上の高次項成分は小さいので、実用上十分精度のある補正を行うことができる。
【００１８】
２次以下の項数は８項であるため、ボクセル数は原理的に８以上であればよく、例えば３×３×３（＝２７）とすることができる。この場合、上記３次元ＭＲＳＩ法における反復ループ数９回で静磁場計測を行うことができる。
【００１９】
更に、本発明の好適な態様として、３次元ＲＳＩ法において、前記計測制御手段は、前記リードアウト方向のマトリクス数を残りの２方向よりも大きく設定する。これにより静磁場計測時間を実質的に延長することなく、スペクトル方向に高精度な情報を得ることができる。
【００２０】
【発明の実施の形態】
以下、本発明を実施例に基づき詳細に説明する。図４は本発明が適応されるＭＲＩ装置の概略構成図であり、この装置は、被検体４０１内部に一様な静磁場Ｂ０を発生させるための静磁場発生磁気回路４０２、直交するｘ，ｙおよびｚの３方向に強度が線形に変化する磁場勾配を与えるための傾斜磁場発生系４０３，被検体４０１に高周波磁場を発生する送信系４０４，被検体から生じる核磁気共鳴信号を検出するための検出系４０５，検出された信号を処理，記憶するための信号処理系４０６，これら傾斜磁場発生系４０３，送信系４０４，検出系４０５を制御するシーケンサ４０７及び装置全体の制御及び信号処理系における各種演算を行うコンピュータ４０８を備え、更にコンピュータ４０８に指令を与える操作部４２１を備える。
【００２１】
静磁場発生磁気回路４０２は、図示しないが均一な静磁場を発生するための電磁石または永久磁石と、静磁場の不均一性を補正するためのシムコイル４３０及びシム電源４３１とを備えている。シムコイル４３０は、例えばｚ２，ｘｚ，ｘｙ，ｙｚ，ｘ２−ｙ２等の２次コイルやｚ３，ｚ２ｘ，ｚ２ｙ等の３次コイルを組合せて用いる。１次項の補正は、ｘ，ｙ，ｚ方向の線形傾斜磁場を発生する傾斜磁場コイル４０９を用い、本来の傾斜磁場発生電流にシム補正電流を重畳して供給する。本発明においてシミング手段とは、これらシムコイル４３０及び傾斜磁場コイル４０９を含む。図６はシムコイルによって発生する磁場分布の一例を示したもので、ｙ−ｚ面内分布を、正側を実線で、負側を破線で等高線表示している。
【００２２】
次に装置の動作の概要を説明する。送信系４０４においてシンセサイザ４１１により発生させた高周波を変調器４１２で変調し電力増幅器４１３で増幅し、送信コイル４１４ａに供給することにより被検体４０１の内部に高周波磁場を発生させ、核スピンを励起させる。通常は¹Ｈを対象とするが、³¹Ｐ，¹²Ｃ等、核スピンを有する他の原子核を対象とすることもある。
【００２３】
この際、被検体から生じる核磁気共鳴信号に位置情報を付与するために傾斜磁場コイル４０９により傾斜磁場Ｇｘ，Ｇｙ，Ｇｚを発生する。３軸方向の傾斜磁場コイル４０９は、それぞれ電源４１０から電流の供給を受け、高周波磁場の照射とともにシーケンサが制御する所定のパルスシーケンスに従って駆動される。
【００２４】
被検体４０１から放出される核磁気共鳴信号は検出系４０５において受信コイル４１４ｂにより受信され、増幅器４１５を通った後、検波器４１６で直交位相検波され、シーケンサ４０７からの命令によるタイミングでＡ／Ｄ変換器４１７を経てコンピュータ４０８へ入力される。高周波コイル４１４ａ，４１４ｂは送受信両用の１つのコイルであってよく、別々でもよい。コンピュータ４０８は信号処理後、前記核スピンの密度分布，緩和時間分布，スペクトル分布等に対応する画像をＣＲＴ等のディスプレイ４２８に表示する。計算途中のデータあるいは最終データはメモリ４２５に収納される。傾斜磁場発生系４０３，送信系４０４，検出系４０５は全てシーケンサ４０７によって制御され、このシーケンサ４０７はコンピュータ４０８によって制御される。
【００２５】
本発明はこのような構成におけるＭＲＩ装置の静磁場の不均一性（静磁場分布）、特に生体に起因する静磁場不均一の計測とそれに基づくシムコイルの制御に関するものであり、１）被検体が置かれた状態での静磁場分布の計測工程と、２）計測された静磁場分布をシムコイルが発生する磁場分布で展開する工程と、３）シムコイルに供給する補正電流を求める工程とからなる。
【００２６】
１）静磁場分布計測工程
静磁場分布計測工程は、高速ＭＲＳＩ法によるパルスシーケンスを採用し、これによってボクセル毎のスペクトルを求める。ここでは一例として図２に示すように被検体頭部２１の３×３×３（ボクセル数＝２７）のマトリクス２２についてスペクトルを得る場合を説明する。まず、一般的な領域選択性ＲＦパルスと拡散傾斜磁場の組合せを複数回印加して、関心領域外を予備飽和（プレサチュレーション）し、関心領域外からの信号の発生を抑制する。続いて図１に示す高速ＭＲＳＩ法を実施する。
【００２７】
この実施例はグラディエントエコー（ＧｒＥ）法を基本とする高速ＭＲＳＩ法を採用しており、関心領域を選択的に励起するＲＦパルス１１を傾斜磁場（ここではｚ方向の傾斜磁場）１３とともに印加し、次に２方向（ここではｚ及びｙ方向）の位相エンコード傾斜磁場１４，１６を印加し、リードアウト方向（ｘ方向）の傾斜磁場１８の反転を繰り返しながら、エコー信号１９を計測する。
【００２８】
ここでＲＦパルス１１のフリップ角αは９０゜でもよいが、生体シミングを短時間で行うために、９０゜より小さいことが好ましい。一例として（９０゜×ＴＲ）／Ｔ１とする。ここでＴ１は、対象とする組織のおおよそのＴ１値であり、繰り返し時間ＴＲとしてＴ１より長い値を用いる場合は、Ｔ１／ＴＲ＝１に置換する。小フリップ角のパルスを用いることにより信号強度の低下を抑えて繰り返し時間ＴＲを短縮できる。フリップ角のとり方については、上述のように一定としてもよいが、励起毎に順次増加させててもよく、その場合発生する横磁化成分を一定にすることができる。
【００２９】
このようなシーケンスを繰り返し時間ＴＲで繰り返し、２方向に位相エンコードループを実行することにより、ｘ軸，ｙ軸，ｚ軸及び時間（δ）軸の４次元データを得る。シーケンスの繰り返しにおいて残留横磁化のコヒーレンスを除去するため、リードアウト方向にはスポイラー傾斜磁場２１を付加し、位相エンコードの累積を防ぐために位相方向にはリワインド傾斜磁場２０を付加する。
【００３０】
位相エンコード方向（ｙ，ｚ方向）のステップ数はその方向のボクセル数に対応し、ここでは３ステップずつのエンコードを反復する。従って３×３回の繰り返しで計測を終える。リードアウト方向のサンプリング点数は、その方向のボクセル数に対応し、ここではサンプリング点数は３点とする。リードアウト方向の傾斜磁場の反転回数は、時間軸方向のマトリックス数の１／２となり、数１０程度とすることができる。
【００３１】
このシーケンスによって計測されたデータのｋ空間配置を図３に示す。図３は、特定の位相ステップｋｙ，ｋｚについて、ｋｘ−ｋδ平面の軌跡を描いており（ｋδ軸は時間軸）、１つのエコー３１について３つのサンプリング点３２，３３，３４でサンプリングされていることを示す。このような計測データにｋｘ，ｋｙ，ｋｚ，ｋδについての４次元ＦＦＴを施し、図２に示すようにボクセル毎のプロトンスペクトルを得る。各ボクセルのスペクトルから水プロトンのピーク周波数２３を得、この周波数から静磁場分布を得る。水プロトンの共鳴周波数を用いることにより、生体内の概ね全てのボクセルに亘り静磁場強度を計測できる。
【００３２】
一例として、リードアウト方向の傾斜磁場を周期６ｍｓの矩形波で２６回反復して印加し、５２のエコーを取得したとする。一般にスペクトル計測帯域Ｌδはｋδ方向のデータ間隔を△ｔ（図３）とするとき、Ｌδ＝１／（△ｔ）の関係があるので、ここでは１／（６ｍｓ）＝１．６７Ｈｚとなる。これは１．５ＴのＭＲＩでは２．６ｐｐｍに相当し、生体内に存在する各種代謝物プロトンのスペクトル帯域（５ｐｐｍ）よりも狭いため、計測帯域外部のピークが帯域内に折り返してくるが、通常の代謝物の強度は水よりも３ないし４桁小さいため、水のピークの同定と位置の検出には障害にならない。皮下脂肪等は水に匹敵する場合もあるが、脂肪のピーク（例えば図２の２４）は水とは隔たっているためこれも障害にはならない。一般に代謝物の定量を行おうとすると折り返しは同定と定量の障害になるが、ここでは大量に存在する水の周波数を計測するだけでよいことが利点となる。
【００３３】
尚、ボクセルの静磁場が２．６ｐｐｍ以上基準値よりもずれていた場合は、水ピークが２回以上折り返されるため真の周波数が判定困難になるが、通常の撮影条件ではこのような大きなずれは生じにくい。また、リードアウト傾斜磁場の反転周期を６ｍｓよりも短くし、データ間隔Δｔを短く設定しておけば折り返しは回避できる。ＥＰＩに対応した高磁場ＭＲＩ装置であれば反転周期を２ｍｓ程度まで短くすることは可能である。
【００３４】
また、図３では簡単のため傾斜磁場の正負のエコーを個別に処理して個別のスペクトルを得るもとのしているが、位相補正等を施した後、合わせて処理してもよく、その場合は△ｔは図３よりも短く（△ｔ／２）なり、計測帯域Ｌδは拡張される。
【００３５】
スペクトルの周波数分解能△ｆは計測時間Ｔと△ｆ＝１／Ｔの関係があり、前掲の例では、Ｔ＝１５６ｍｓ（＝６ｍｓ×２６）、△ｆ＝６．４Ｈｚとなる。これは１．５ＴのＭＲＩでは０．ｌｐｐｍに相当する。従って、静磁場の計測精度も最高で０．ｌｐｐｍとなる。ＥＰＩで顕著なアーチファクトが発生しない静磁場均一度は約０．３ｐｐｍ以下であるため、計測精度としては十分である。
【００３６】
尚、上記実施例ではリードアウト方向のサンプリング点数を３点として説明したが、サンプリング点数は計測時間を延長しない範囲で大きくとることができ、その場合オーバーサンプリングしたデータの平均値を記録することは当然可能である。
【００３７】
水プロトンのピーク周波数は、上述のように４Ｄ−ＦＦＴによって得られたボクセル毎のプロトンスペクトルから計算機により自動検出する。ピーク検出方法としては、標準となる水の位置を中心としてローレンツ曲線をあてはめ、その位置，幅，高さを微調整する方法がある。また、必要に応じて位相補正を併用する。
【００３８】
次にボクセル毎に水のピーク位置から静磁場磁場強度を式（２）により計算する。
【数２】

式中δｗ（ｘ，ｙ，ｚ）は、基準位置からの水ピークのシフト（ｐｐｍ）,Ｂｅ（ｘ，ｙ，ｚ）は静磁場不均一である。
【００３９】
あるボクセルの水ピークの基準位置（４．７ｐｐｍ）からのずれδｗは、そのボクセル内を平均した静磁場の不均一を示すので、各ボクセルについて静磁場強度を求めることにより関心領域全体について静磁場分布をマッピングすることができる。ただし、ボクセル内の水の分布が片寄っている場合は、ボクセル内の水の分布で重み付けした平均となる。
【００４０】
２）計測された静磁場分布をシムコイルが発生する磁場分布で展開する工程
この工程では、先に得られた３×３×３個のボクセルに亘る静磁場分布を球面調和関数で展開する。ＭＲＩでは、多くの場合シムコイルは概略球面調和関数状の磁場を発生するように設計されているので、球面調和関数で展開することにより、その係数から直ちにシム電流を求めることができる。
【００４１】
球面調和関数は表１に示すような関数形からなるが、本実施例では２次以下の項のみを用いて静磁場分布を近似する。一般に３次以上の高次項成分は小さく、実用的には２次以下の項のみで十分静磁場均一化が達成されるからである。表１関数形のうち、１次項はｘ，ｙ，ｚの３項、２次項はｚ²，ｚｘ，ｘ²−ｙ²，ｘｙ，ｚｙの５項であり、これらの合計８頃の係数を求める。
【表１】

各調和関数の係数の決定には例えば勾配法等を用い、式（３）の残差Ｉを最小化する。
【数３】

式中、Ｐｉ（ｘ，ｙ，ｚ）は球面調和関数、Ｃｊはその係数である。ｖに関する総和は、均一化を行う体積内（ここでは２７のボクセル）の各座標（ｘ，ｙ，ｚ）に亘って行う。
【００４２】
ここで計測体積の中心がシムコイルの中心と一致している場合は、係数の決定は容易である。一致していない場合は、計測した静磁場分布Ｂｅ（ｘ，ｙ，ｚ）と各シムコイルの生成する磁場分布Ｐｉ（ｘ，ｙ，ｚ）の座標系が一致するよう変換処理を行ってから係数を決定する。
【００４３】
３）シム電流を求める工程
この工程では、前工程で得られた球面調和関数の係数からシムコイルへ流す補正電流を求める。補正電流は理想的には計測した静磁場分布と振幅が等しく、向きが反対の補正磁場分布を発生する。前述したように、シムコイルが概略球面調和関数状の磁場を発生するように設計されている場合には、球面調和関数の係数から直ちにシム電流を求めることができる。この場合、シムのコイル毎の固有の磁場発生効率を考慮する。
【００４４】
別法として、各シムコイルの生成する磁場分布（シム特性）から、式（４）に示す行列演算により補正電流を求めてもよい。
【数４】

式中、Ａはシム特性行列（Ａｊｋ＝δＢｊ／δＩｋ）を表し、要素は第ｋチャンネルシム電流の微小変化に対する第ｊ画素の静磁場変化で表される。△Ｉはシム電流ベクトル、Ｂ０は画素毎の静磁場偏倚を１次元に再配列したベクトル、ＡｔはＡの転置行列である。シム特性は、予め水などの均質な試料を用いて、シムコイルに流す単位電流当りの静磁場分布の変化を測定することにより、求めることができ、行列或いは調和関数で展開した展開関数の形式でメモリ内に記憶される。
【００４５】
この方法はシムコイルが球面調和関数とは異なる磁場、あるいは複数の球面調和関数の合成磁場を発生する場合にも適用できる。
【００４６】
以上の３つの工程は、本撮像に先立つプリスキャンとして行われ、これら工程で得たシム電流のうち、１次シム値は傾斜磁場のオフセットとして設定し、２次シム値は２次シムコイルへ設定する。しかる後に本撮像を実行する。
【００４７】
この方法によれば、きわめて短時間に生体シミングを実行することができ、ＥＰＩ法やスペクトロスコピックイメージング法など高い静磁場均一性が要求される撮像において高画質画像を得ることができる。一例として第１の工程におけるボクセル数を３×３×３（＝２７），ＲＦパルスのフリップ角１０゜，シーケンスの繰り返し時間ＴＲ＝１６０ｍｓとすると、１６０ｍｓ×９＝１．４４秒の計測時間＋演算時間で静磁場補正を行うことができる。従って、シム電流値の計算を含む全シミング工程を数秒程度で完了できる。
【００４８】
尚、以上の実施例では、第１工程におけるボクセル数は、静磁場分布を展開する場合の展開項として２次項まで用いることを前提として、（３×３×３）の場合を説明したが、既に述べたようにボクセル数は位相エンコードステップ数及び周波数エンコード方向のサンプリング点数によって決まり、これらを変えることにより関心領域に合せた任意のボクセル数、形状とすることができる。
【００４９】
また展開項として３次，４次等より高次の項を含めることも可能であり、これにより高精度なシミングが可能になる。その場合、項の数に対応して体積内の画素マトリクス数を増やす必要があるが、リードアウト方向の画像マトリクス数は計測時間を延長せずに増加させることができるので、この方向にマトリクスを大きくとり、３次以上の展開項を用いることは有効である。従って生体の体軸方向等、高次の不均一が予想される方向をリードアウト方向に選ぶのが有効である。
【００５０】
また、位相エンコード数を４以上とし、マトリクスを位相方向にも拡大すれば、３次以上の展開項も使用できるようになり、空間的により高精度な均一化が可能になることは言うまでもない。この場合位相エンコードステップ数と共に計測時間は増大するが、本発明で採用する高速ＭＲＳＩ法の時間的優位性はエンコードステップ数が大きいほど顕著になる。
【００５１】
更に上記実施例では静磁場分布測定のためのシーケンスとして図１に示すグラディエントエコー（ＧｒＥ）型の高速ＭＲＳＩ法を採用したが、図５に示すＳＥ型の高速ＭＲＳＩ法を採用してもよい。図５のシーケンスは高周波パルス５１を照射し、ＴＥ／２経過後にスピンを反転させる高周波パルス５１０を照射している点が異なり、２方向に位相エンコード５４，５５を付与すること、リードアウト方向に反転する傾斜磁場５８を印加すること、繰り返し時間の最後にリワインド５０１及びスポイラー５１２をそれぞれ付加することは図１の場合と同様である。
【００５２】
ＧｒＥの型のシーケンスとＳＥ型のシーケンスとを比較すると、前者は低周波領域の信号が犠牲になるので、スペクトルのベースラインのうねりが生じる。但し、水ピークの検出には大きな障害にはならない。ＳＥ型は、計測時間はＧｒＥ型よりも長くなるが、エコー中心のデータが取得できるので、ベースラインのうねりを防止できる。
【００５３】
また、ＳＥ型との特長として、公知のＴ２を用いるスペクトル編集が可能になる。即ち、エコー時間ＴＥを数１０ｍｓと大きく設定することにより、脂肪等の短Ｔ２物質のスペクトルを消去したスペクトルを得ることができる。スペクトル線の半値幅ν１／２はＴ２とν１／２＝１／（πＴ２）の関係があるため、短Ｔ２物質のピークはブロードであり、短Ｔ２物質を消去したスペクトルはシャープになる。従ってピークの位置を高精度に検出できる。
【００５４】
【発明の効果】
以上説明したように、本発明によれば生体シミングを行うに際し、静磁場不均一性を測定するために高速ＭＲＳＩ法によるシーケンスを小エンコードステップで実行するとともに、得られた静磁場分布を低次項の調和関数で展開することにより、分布被写体に応じた体積内静磁場均一化を短時間で達成でき、ＥＰＩ法やＥＰＩ法をべースとした撮影方法の画質を向上させることができる。
【図面の簡単な説明】
【図１】本発明の静磁場計測シーケンスの一実施例を示す図。
【図２】スペクトルによる静磁場分布計測の概念を示す図。
【図３】高速ＭＲＳＩ法のｋ空間軌跡を示す図。
【図４】本発明が適用されるＭＲＩ装置の全体の構成を示す図。
【図５】本発明の静磁場計測シーケンスの他の実施例を示す図。
【図６】シムコイルの磁場分布を示す図。
【符号の説明】
４０１ 被写体
４１４ａ 送信ＲＦコイル
４１４ｂ 検出ＲＦコイル
４３０ シムコイル
４３１ シム電源[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for homogenizing a static magnetic field in a magnetic resonance imaging (hereinafter referred to as MRI) apparatus, and more particularly to a method for rapidly homogenizing a static magnetic field in a predetermined volume portion with a subject placed.
[0002]
[Prior art]
In the 2D or 3D Fourier transform method, which is a typical image construction method of MRI, a subject is placed in a static magnetic field that is spatially uniform and directed in a certain direction (this is the z direction), and the subject's nucleus By applying a high-frequency pulse to the horizontal plane (x, y plane) and then applying a predetermined combination of gradient magnetic fields, a space is generated in the phase of the nuclear spin (reference direction, for example, the azimuth angle from the y-axis direction). A signal is measured by generating a distribution according to the target position.
[0003]
However, when the static magnetic field is not uniform, phase rotation occurs due to this, and thereby a false image, positional distortion, etc. occur, and the image quality deteriorates. The static magnetic field inhomogeneity is first caused by the device limitations of the static magnetic field generating magnet. Furthermore, since the living body itself has a slightly different magnetic permeability for each tissue, the static magnetic field inside the subject is distributed.
[0004]
In MRI, shimming (static magnetic field homogenization) is important in reducing such artifacts, and in particular, correction of non-uniform static magnetic field caused by a living body (hereinafter referred to as living body shimming) is performed for each subject. Because it is necessary, the processing must be completed in a short time.
[0005]
In general, in living body shimming, a static magnetic field distribution induced by a magnetic permeability distribution in a subject is measured in advance by an MRI technique, a correction shim current is calculated based on the static magnetic field distribution, and this current is passed through a shim coil to cause nonuniformity. The static magnetic field distribution is flattened by generating a reverse magnetic field. This imaging is performed under this shimming.
[0006]
As a method for measuring a static magnetic field distribution in a living body, a method for obtaining a static magnetic field distribution in a subject from two complex images having different magnetization development times (JP-A-4-288136 and US Pat. No. 5,168,232). In addition, Grover et al. Have proposed a method of obtaining a static magnetic field distribution separately from a chemical shift by measuring an image while suppressing a specific chemical shift (Japanese Patent Publication No. 6-44904). Each of these methods makes the phase distribution due to the static magnetic field distribution occur in the signal by making the generation time of the RF spin echo different from the generation time of the gradient magnetic field echo, and obtains the static magnetic field distribution from this phase distribution. is there.
[0007]
Since the static magnetic field distribution measurement is performed for each patient as a preliminary imaging prior to the main imaging, a method for obtaining a phase distribution image with a single excitation (for example, “dynamic magnetic field change in the lung and heart” is used as a method of speeding up this. And Dynamic Field Changes and EPI Geometric Distortions in the Chest and Heart (S. Kanayama, S. Kuhara and K. Satoh, Proceedings of the Society of Magnetic Resonance, 746 (1995)) Yes.
[0008]
As a method for obtaining a static magnetic field distribution from these phase distributions, if an EPI method or an FSE method is used, the measurement of the static magnetic field distribution in the slice can be performed in a short time of about several seconds, for example. In addition, the image matrix can be made relatively large, like a normal image, for example, about 34 to 128. Therefore, it can be said to be an effective method for measuring the static magnetic field distribution in the slice.
[0009]
However, in order to obtain a static magnetic field distribution in a three-dimensional volume, it is necessary to repeat phase encoding in the thickness direction. If the matrix in the thickness direction is made as large as a normal image, the advantage of short-time measurement is lost. It will be broken. Conversely, if the matrix in the thickness direction is reduced, the significance of increasing only the matrix in the slice is diminished. Another problem associated with using phase is the problem of phase unwrapping. This is a problem of how to find the main value when a phase difference of 2π radians or more occurs between pixels, and complicated processing is required.
[0010]
As a static magnetic field distribution measurement method without using a phase, there is a method using a chemical shift. This is a 3D / 4D-chemical shift imaging (CSI) method that measures the spectrum of multi-pixels or voxels, and obtains the local magnetic field strength directly from the frequency of the spectral lines of specific molecules such as water protons. is there. From the frequency of the spectrum, the local static magnetic field strength is obtained by equation (1).
[Expression 1]

This static magnetic field distribution measurement method is applied to correct a spectral position other than water based on the static magnetic field intensity and static magnetic field uniformity of a pixel.
[0011]
[Problems to be solved by the invention]
However, when the 3D / 4D-CSI method is applied as a preliminary imaging for biological shimming, there is a drawback that it takes a long time for measurement. For example, in the measurement of a three-dimensional volume, the repetition loop of phase encoding is triple, so that excitation of the third power of the number of matrices is required. As an example, if the repetition time (TR) is 1 second and the matrix is 16 × 16 × 16, 1 second × 16 × 16 × 16 = about 68 minutes is required.
[0012]
[Means for Solving the Problems]
In order to solve the above problems, in the present invention, a three-dimensional high-speed MR spectroscopic imaging method (hereinafter referred to as MRSI) is used as a static magnetic field distribution measurement method in living body shimming including static magnetic field distribution measurement and shim current determination. (P. Mansfield's paper "Spatial Mapping of the Chemical Shift in NMR", Magnetic Resonance in Medicine 1, 370-386 (1984), Matsui et al. (See High-Speed Spatially Resolved High-Resolution NMR Spectroscoy, J. Am. Chem. Soc., 107, 2817-2818 (1985)). Further, in determining the shim current, the development of the static magnetic field distribution is limited to the second order term. This provides a practical and quick method for homogenizing a static magnetic field.
[0013]
That is, the magnetic resonance imaging apparatus of the present invention comprises a static magnetic field generating means for generating a static magnetic field, a shimming means for correcting the static magnetic field inhomogeneity for each spatial component, a slice direction of the static magnetic field space, and a phase encoding. A gradient magnetic field generating means for generating a gradient magnetic field of each of a slice gradient magnetic field, a phase encoding gradient magnetic field, and a readout gradient magnetic field in each of a direction and a readout direction, and an atomic composition of an object arranged in the static magnetic field space From a high-frequency magnetic field generating means for generating a high-frequency magnetic field for inducing nuclear magnetic resonance in a nuclear spin, a receiving means for receiving an echo signal generated by nuclear magnetic resonance from the subject, and from the received nuclear magnetic resonance signal Signal processing means for reconstructing the image of the subject, the gradient magnetic field generating means and the high-frequency magnet based on a predetermined pulse sequence And a measurement control means for controlling said signal processing means and generating means and the receiving means,
The measurement control means excites a desired volume portion of the subject by irradiation with the high-frequency magnetic field, and then an echo signal representing the static magnetic field inhomogeneity for the volume portion by a predetermined static magnetic field inhomogeneity detection pulse sequence. Get
The signal processing means obtains a static magnetic field distribution of the volume portion from an echo signal representing the static magnetic field inhomogeneity, expands the static magnetic field distribution with a magnetic field distribution for each spatial component generated by the shimming means, and generates the static magnetic field. Find the optimum value of the current to flow for each spatial component of the shimming means so as to make the distribution uniform,
In the magnetic resonance imaging apparatus in which the shimming means flows the current of the optimum value for each spatial component,
The static magnetic field inhomogeneity detection pulse sequence repeats a basic sequence of applying a slice encode gradient magnetic field in the slice direction, a phase encode gradient magnetic field in the phase encode direction, and a gradient magnetic field that continuously inverts in the readout direction.
The signal processing means includes The readout direction gradient magnetic field has the same polarity and is acquired at the same interval in the time axis direction. A spectrum for each voxel of the subject is obtained from an echo signal representing the static magnetic field inhomogeneity, and a static magnetic field distribution for each voxel is obtained by converting a resonance frequency of the specific peak into a static magnetic field intensity.
[0014]
The three-dimensional MRSI method is a high-speed spectroscopic imaging method that reduces the repetitive loop by one dimension by continuously reversing the gradient magnetic field in the readout direction and superimposing the spatial information and the spectral information. A multi-voxel spectrum can be acquired.
[0015]
It is preferable to use the resonance frequency of water protons as the specific spectral peak. Thereby, the static magnetic field intensity can be measured over almost all voxels in the living body. Further, even when the repetition time (TR) is shortened for shortening the time, a high signal can be obtained by using a pulse with a low flip angle as the high frequency magnetic field. Furthermore, when a sequence based on the spin echo (SE) method is adopted as the three-dimensional MRSI method and the TE is set to be long, a sharp spectrum in which a signal from bound water or the like is eliminated is obtained using the difference in T2. This improves the reading accuracy of the resonance frequency.
[0016]
In a preferred embodiment of the present invention, The signal processing means expands the static magnetic field distribution for each voxel using the first and second order terms of a spherical harmonic function, finds the optimum current value for each spatial component, and the shimming means It has at least eight spatial components that generate the first-order term and the second-order term of the harmonic function, and the optimum current corresponding to each spatial component is supplied. In the present invention, as a shimming means, a gradient magnetic field coil having a primary term magnetic field distribution and a shim coil having a higher order magnetic field distribution of approximately a second order term are used.
[0017]
By approximating the static magnetic field distribution using only the second and lower terms, the expansion term can be determined from the static magnetic field measurement data of a small number of voxels, and the static magnetic field measurement can be performed within a practical short time. In addition, by setting the second order or less, it is possible to avoid the difficulty of manufacturing a shim coil that generates only a specific high order term as shimming means, and generally the third order or higher order component is small, so that the correction is sufficiently accurate in practice. It can be performed.
[0018]
Since the number of terms of the second order or less is 8, the number of voxels may be 8 or more in principle, for example, 3 × 3 × 3 (= 27). In this case, the static magnetic field measurement can be performed with 9 iteration loops in the three-dimensional MRSI method.
[0019]
Furthermore, as a preferred embodiment of the present invention, in the three-dimensional RSI method, The measurement control means sets the number of matrices in the readout direction to be larger than the remaining two directions. This makes it possible to obtain highly accurate information in the spectral direction without substantially extending the static magnetic field measurement time.
[0020]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described in detail based on examples. FIG. 4 is a schematic configuration diagram of an MRI apparatus to which the present invention is applied. This apparatus includes a static magnetic field generating magnetic circuit 402 for generating a uniform static magnetic field B0 inside a subject 401, orthogonal x, y And a gradient magnetic field generation system 403 for providing a magnetic field gradient whose intensity changes linearly in the three directions z and z, a transmission system 404 for generating a high-frequency magnetic field in the subject 401, and a nuclear magnetic resonance signal generated from the subject A detection system 405, a signal processing system 406 for processing and storing detected signals, a gradient magnetic field generation system 403, a transmission system 404, a sequencer 407 for controlling the detection system 405, and various controls in the entire apparatus and signal processing system A computer 408 that performs calculation is provided, and an operation unit 421 that gives commands to the computer 408 is further provided.
[0021]
Although not shown, the static magnetic field generating magnetic circuit 402 includes an electromagnet or a permanent magnet for generating a uniform static magnetic field, a shim coil 430 and a shim power source 431 for correcting nonuniformity of the static magnetic field. The shim coil 430 uses, for example, a secondary coil such as z2, xz, xy, yz, and x2-y2 and a tertiary coil such as z3, z2x, and z2y. For the correction of the first-order term, a gradient magnetic field coil 409 that generates a linear gradient magnetic field in the x, y, and z directions is used, and a shim correction current is superimposed and supplied on the original gradient magnetic field generation current. In the present invention, the shimming means includes the shim coil 430 and the gradient magnetic field coil 409. FIG. 6 shows an example of the magnetic field distribution generated by the shim coil, and the yz in-plane distribution is indicated by contour lines with a solid line on the positive side and with a broken line on the negative side.
[0022]
Next, an outline of the operation of the apparatus will be described. In the transmission system 404, the high frequency generated by the synthesizer 411 is modulated by the modulator 412, amplified by the power amplifier 413, and supplied to the transmission coil 414 a to generate a high frequency magnetic field inside the subject 401 to excite nuclear spin. . Normally ¹ For H, ³¹ P, ¹² Other atomic nuclei having a nuclear spin such as C may be targeted.
[0023]
At this time, gradient magnetic fields Gx, Gy, and Gz are generated by the gradient coil 409 in order to give position information to the nuclear magnetic resonance signal generated from the subject. The gradient magnetic field coils 409 in the three-axis directions are each supplied with current from the power supply 410, and are driven according to a predetermined pulse sequence controlled by the sequencer along with the irradiation of the high frequency magnetic field.
[0024]
The nuclear magnetic resonance signal emitted from the subject 401 is received by the receiving coil 414 b in the detection system 405, passes through the amplifier 415, is quadrature detected by the detector 416, and is A / D at the timing according to the command from the sequencer 407. The data is input to the computer 408 via the converter 417. The high frequency coils 414a and 414b may be one coil for both transmission and reception, or may be separate. After the signal processing, the computer 408 displays an image corresponding to the nuclear spin density distribution, relaxation time distribution, spectral distribution, and the like on a display 428 such as a CRT. Data being calculated or final data is stored in the memory 425. The gradient magnetic field generation system 403, transmission system 404, and detection system 405 are all controlled by a sequencer 407, which is controlled by a computer 408.
[0025]
The present invention relates to static magnetic field inhomogeneity (static magnetic field distribution) of an MRI apparatus having such a configuration, particularly to measurement of static magnetic field inhomogeneity caused by a living body and control of a shim coil based on the measurement. It comprises a step of measuring a static magnetic field distribution in a placed state, 2) a step of developing the measured static magnetic field distribution with a magnetic field distribution generated by the shim coil, and 3) a step of obtaining a correction current to be supplied to the shim coil.
[0026]
1) Static magnetic field distribution measurement process
The static magnetic field distribution measurement step employs a pulse sequence based on the high-speed MRSI method, and thereby obtains a spectrum for each voxel. Here, as an example, a case will be described in which a spectrum is obtained for a 3 × 3 × 3 (number of voxels = 27) matrix 22 of the subject's head 21 as shown in FIG. First, a combination of a general region-selective RF pulse and a diffusion gradient magnetic field is applied a plurality of times to presaturate outside the region of interest (presaturation), thereby suppressing the generation of signals from outside the region of interest. Subsequently, the high-speed MRSI method shown in FIG. 1 is performed.
[0027]
This embodiment employs a high-speed MRSI method based on a gradient echo (GrE) method, and applies an RF pulse 11 for selectively exciting a region of interest together with a gradient magnetic field (here, a gradient magnetic field in the z direction) 13. Then, the phase encode gradient magnetic fields 14 and 16 in two directions (here, z and y directions) are applied, and the echo signal 19 is measured while repeating inversion of the gradient magnetic field 18 in the readout direction (x direction).
[0028]
Here, the flip angle α of the RF pulse 11 may be 90 °, but is preferably smaller than 90 ° in order to perform living body shimming in a short time. As an example, (90 ° × TR) / T1. Here, T1 is an approximate T1 value of the target tissue, and when a value longer than T1 is used as the repetition time TR, it is replaced with T1 / TR = 1. By using a pulse with a small flip angle, it is possible to suppress a decrease in signal intensity and shorten the repetition time TR. The method of taking the flip angle may be constant as described above, but may be increased sequentially for each excitation, and the transverse magnetization component generated in that case can be made constant.
[0029]
By repeating such a sequence at a repetition time TR and executing a phase encoding loop in two directions, four-dimensional data of the x-axis, y-axis, z-axis, and time (δ) axis are obtained. In order to remove the coherence of residual transverse magnetization in the repetition of the sequence, a spoiler gradient magnetic field 21 is added in the readout direction, and a rewind gradient magnetic field 20 is added in the phase direction to prevent accumulation of phase encoding.
[0030]
The number of steps in the phase encoding direction (y, z direction) corresponds to the number of voxels in that direction, and here, the encoding is repeated every three steps. Therefore, the measurement is completed by repeating 3 × 3 times. The number of sampling points in the lead-out direction corresponds to the number of voxels in that direction, and here the number of sampling points is three. The number of inversions of the gradient magnetic field in the readout direction is ½ of the number of matrices in the time axis direction, and can be about several tens.
[0031]
FIG. 3 shows the k-space arrangement of data measured by this sequence. FIG. 3 depicts the locus of the kx-kδ plane for a specific phase step ky, kz (kδ axis is a time axis), and one echo 31 is sampled at three sampling points 32, 33, 34. It shows that. Such measurement data is subjected to a four-dimensional FFT for kx, ky, kz, and kδ to obtain a proton spectrum for each voxel as shown in FIG. The water proton peak frequency 23 is obtained from the spectrum of each voxel, and the static magnetic field distribution is obtained from this frequency. By using the resonance frequency of water protons, the static magnetic field strength can be measured over almost all voxels in the living body.
[0032]
As an example, assume that a gradient magnetic field in the readout direction is applied 26 times with a rectangular wave having a period of 6 ms, and 52 echoes are acquired. In general, the spectrum measurement band Lδ has a relationship of Lδ = 1 / (Δt) when the data interval in the kδ direction is Δt (FIG. 3), and therefore, here, 1 / (6 ms) = 1.67 Hz. This is equivalent to 2.6 ppm in 1.5T MRI, and is narrower than the spectrum band (5 ppm) of various metabolite protons present in the living body. Since the metabolite intensity of is 3 to 4 orders of magnitude less than water, it does not interfere with the identification and location of water peaks. Subcutaneous fat or the like may be comparable to water, but since the fat peak (for example, 24 in FIG. 2) is separated from water, this is not an obstacle. In general, quantification of metabolites is an obstacle to identification and quantification, but here it is advantageous that only the frequency of water present in large quantities needs to be measured.
[0033]
If the voxel's static magnetic field deviates by more than 2.6 ppm from the reference value, the water peak will be folded twice or more, making it difficult to determine the true frequency. Is unlikely to occur. Further, if the inversion period of the readout gradient magnetic field is set shorter than 6 ms and the data interval Δt is set short, folding can be avoided. In the case of a high magnetic field MRI apparatus compatible with EPI, the inversion period can be shortened to about 2 ms.
[0034]
Further, in FIG. 3, for the sake of simplicity, the positive and negative echoes of the gradient magnetic field are individually processed to obtain individual spectra. However, after performing phase correction or the like, they may be processed together. In this case, Δt is shorter than that in FIG. 3 (Δt / 2), and the measurement band Lδ is expanded.
[0035]
The frequency resolution Δf of the spectrum has a relationship of the measurement time T and Δf = 1 / T. In the above example, T = 156 ms (= 6 ms × 26) and Δf = 6.4 Hz. This is 0.4 for 1.5T MRI. Corresponds to 1 ppm. Therefore, the measurement accuracy of the static magnetic field is 0. 1 ppm. Since the static magnetic field uniformity at which no significant artifacts occur in EPI is about 0.3 ppm or less, the measurement accuracy is sufficient.
[0036]
In the above embodiment, the number of sampling points in the lead-out direction has been described as three points. However, the number of sampling points can be increased within a range that does not extend the measurement time, and in that case, the average value of oversampled data can be recorded. Of course it is possible.
[0037]
The peak frequency of the water proton is automatically detected by a computer from the proton spectrum for each voxel obtained by 4D-FFT as described above. As a peak detection method, there is a method of fitting a Lorentz curve centering on a standard water position and finely adjusting the position, width and height. Further, phase correction is used in combination as necessary.
[0038]
Next, the static magnetic field strength is calculated from the peak position of water for each voxel by the equation (2).
[Expression 2]

In the equation, δw (x, y, z) is a shift (ppm) of the water peak from the reference position, and Be (x, y, z) is non-uniform in static magnetic field.
[0039]
The deviation δw from the reference position (4.7 ppm) of the water peak of a certain voxel indicates the non-uniformity of the static magnetic field averaged in the voxel. Therefore, by obtaining the static magnetic field strength for each voxel, the static magnetic field for the entire region of interest is obtained. Distributions can be mapped. However, when the distribution of water in the voxel is offset, the average is weighted by the distribution of water in the voxel.
[0040]
2) The process of developing the measured static magnetic field distribution with the magnetic field distribution generated by the shim coil
In this step, the static magnetic field distribution over the 3 × 3 × 3 voxels obtained previously is developed with a spherical harmonic function. In MRI, in many cases, the shim coil is designed to generate a magnetic field having a substantially spherical harmonic function. Therefore, by developing the spherical harmonic function, the shim current can be obtained immediately from the coefficient.
[0041]
The spherical harmonic function has a function form as shown in Table 1. In this embodiment, the static magnetic field distribution is approximated using only the second-order or lower terms. This is because the higher-order component of the third order or higher is generally small, and practically, the static magnetic field homogenization can be sufficiently achieved only by the second-order or lower term. In Table 1 function form, the primary term is x, y, z 3 terms, the secondary term is z ² , Zx, x ² -Y ² , Xy, and zy, and a total of these coefficients is obtained.
[Table 1]

For example, the gradient method or the like is used to determine the coefficient of each harmonic function, and the residual I in Expression (3) is minimized.
[Equation 3]

In the equation, Pi (x, y, z) is a spherical harmonic function, and Cj is its coefficient. The summation regarding v is performed over each coordinate (x, y, z) within the volume to be equalized (here, 27 voxels).
[0042]
Here, when the center of the measurement volume coincides with the center of the shim coil, the coefficient can be easily determined. If they do not match, the coefficient is calculated after performing conversion processing so that the measured static magnetic field distribution Be (x, y, z) and the coordinate system of the magnetic field distribution Pi (x, y, z) generated by each shim coil match. To decide.
[0043]
3) Step of obtaining shim current
In this step, a correction current to be supplied to the shim coil is obtained from the coefficient of the spherical harmonic function obtained in the previous step. The correction current ideally generates a correction magnetic field distribution having the same amplitude and opposite direction as the measured static magnetic field distribution. As described above, when the shim coil is designed to generate a magnetic field having a substantially spherical harmonic function, the shim current can be obtained immediately from the coefficient of the spherical harmonic function. In this case, the inherent magnetic field generation efficiency for each shim coil is considered.
[0044]
As an alternative method, the correction current may be obtained from the magnetic field distribution (shim characteristics) generated by each shim coil by the matrix calculation shown in Expression (4).
[Expression 4]

In the equation, A represents a shim characteristic matrix (Ajk = δBj / δIk), and an element is represented by a change in the static magnetic field of the jth pixel with respect to a minute change in the kth channel shim current. ΔI is a shim current vector, B0 is a vector in which the static magnetic field deviation for each pixel is rearranged in one dimension, and At is a transposed matrix of A. Shim characteristics can be obtained by measuring changes in the static magnetic field distribution per unit current flowing in the shim coil using a homogeneous sample such as water in advance, in the form of an expansion function developed by a matrix or harmonic function. Stored in memory.
[0045]
This method can also be applied to the case where the shim coil generates a magnetic field different from the spherical harmonic function or a combined magnetic field of a plurality of spherical harmonic functions.
[0046]
The above three steps are performed as a pre-scan prior to the main imaging. Among the shim currents obtained in these steps, the primary shim value is set as an offset of the gradient magnetic field, and the secondary shim value is set to the secondary shim coil. To do. Thereafter, the main imaging is executed.
[0047]
According to this method, living body shimming can be performed in a very short time, and a high-quality image can be obtained in imaging that requires high static magnetic field uniformity, such as the EPI method or spectroscopic imaging method. As an example, assuming that the number of voxels in the first step is 3 × 3 × 3 (= 27), the flip angle of the RF pulse is 10 °, and the sequence repetition time TR = 160 ms, the measurement time of 160 ms × 9 = 1.44 seconds + Static magnetic field correction can be performed in calculation time. Therefore, the entire shimming process including the calculation of the shim current value can be completed in about several seconds.
[0048]
In the above embodiment, the number of voxels in the first step has been described as being (3 × 3 × 3) on the premise that the quadratic term is used as the expansion term when the static magnetic field distribution is expanded. As described above, the number of voxels is determined by the number of phase encoding steps and the number of sampling points in the frequency encoding direction, and by changing these, it is possible to obtain an arbitrary number of voxels and shapes according to the region of interest.
[0049]
Further, it is possible to include higher-order terms than third-order, fourth-order, etc. as expansion terms, which enables highly accurate shimming. In that case, it is necessary to increase the number of pixel matrices in the volume corresponding to the number of terms, but the number of image matrices in the readout direction can be increased without extending the measurement time. It is effective to use a third or higher order expansion term. Therefore, it is effective to select a direction in which higher-order non-uniformity such as the body axis direction of the living body is expected as the lead-out direction.
[0050]
Further, if the number of phase encodes is set to 4 or more and the matrix is expanded also in the phase direction, it is possible to use a third-order or higher-order expansion term, and it is needless to say that spatially highly uniformization is possible. In this case, the measurement time increases with the number of phase encoding steps, but the temporal advantage of the high-speed MRSI method employed in the present invention becomes more significant as the number of encoding steps increases.
[0051]
Further, in the above embodiment, the gradient echo (GrE) type high-speed MRSI method shown in FIG. 1 is adopted as a sequence for measuring the static magnetic field distribution, but the SE type high-speed MRSI method shown in FIG. 5 may be adopted. The sequence shown in FIG. 5 is different in that a high frequency pulse 51 is applied and a high frequency pulse 510 is applied to invert the spin after the lapse of TE / 2, and phase encodings 54 and 55 are applied in two directions, and in the readout direction. The application of the reversing gradient magnetic field 58 and the addition of the rewind 501 and the spoiler 512 at the end of the repetition time are the same as in the case of FIG.
[0052]
Comparing the GrE type sequence with the SE type sequence, the former sacrifices the signal in the low frequency region, resulting in a baseline waviness of the spectrum. However, it is not a major obstacle to the detection of water peaks. The SE type has a longer measurement time than the GrE type, but the echo center data can be acquired, so that the baseline swell can be prevented.
[0053]
Further, as a feature of the SE type, spectrum editing using a known T2 becomes possible. That is, by setting the echo time TE as large as several tens of ms, a spectrum in which the spectrum of a short T2 substance such as fat is eliminated can be obtained. Since the half width ν1 / 2 of the spectrum line has a relationship of T2 and ν1 / 2 = 1 / (πT2), the peak of the short T2 substance is broad, and the spectrum from which the short T2 substance is erased becomes sharp. Therefore, the peak position can be detected with high accuracy.
[0054]
【The invention's effect】
As described above, according to the present invention, when performing living body shimming, a sequence by the high-speed MRSI method is executed in a small encoding step in order to measure the static magnetic field inhomogeneity, and the obtained static magnetic field distribution is expressed by a low-order term By developing with the harmonic function, it is possible to achieve in-volume static magnetic field homogenization corresponding to the distributed subject in a short time, and to improve the image quality of the EPI method and the imaging method based on the EPI method.
[Brief description of the drawings]
FIG. 1 is a diagram showing an example of a static magnetic field measurement sequence of the present invention.
FIG. 2 is a diagram showing the concept of static magnetic field distribution measurement by spectrum.
FIG. 3 is a diagram showing a k-space trajectory of a high-speed MRSI method.
FIG. 4 is a diagram showing the overall configuration of an MRI apparatus to which the present invention is applied.
FIG. 5 is a diagram showing another embodiment of the static magnetic field measurement sequence of the present invention.
FIG. 6 is a diagram showing a magnetic field distribution of shim coils.
[Explanation of symbols]
401 Subject
414a Transmit RF coil
414b Detection RF coil
430 shim coil
431 Shim Power

Claims (3)

Static magnetic field generating means for generating a static magnetic field, shimming means for correcting non-uniformity of the static magnetic field for each spatial component, and slice gradient magnetic fields in the slice direction, phase encoding direction, and readout direction of the static magnetic field space, respectively. A gradient magnetic field generating means for generating a gradient magnetic field, a phase encoding gradient magnetic field, and a readout gradient magnetic field, and for inducing nuclear magnetic resonance in a nuclear spin of an atom composing an object arranged in the static magnetic field space High-frequency magnetic field generating means for generating a high-frequency magnetic field, receiving means for receiving an echo signal generated by nuclear magnetic resonance from the subject, and signal processing for reconstructing the image of the subject from the received nuclear magnetic resonance signal Means, said gradient magnetic field generating means, said high frequency magnetic field generating means, said receiving means, and said signal processing means based on a predetermined pulse sequence And a control for the measuring and controlling unit,
The measurement control means excites a desired volume portion of the subject by irradiation with the high-frequency magnetic field, and then an echo signal representing the static magnetic field inhomogeneity for the volume portion by a predetermined static magnetic field inhomogeneity detection pulse sequence. Get
The signal processing means obtains a static magnetic field distribution of the volume portion from an echo signal representing the static magnetic field inhomogeneity, expands the static magnetic field distribution with a magnetic field distribution for each spatial component generated by the shimming means, and generates the static magnetic field. Find the optimum value of the current to flow for each spatial component of the shimming means so as to make the distribution uniform,
In the magnetic resonance imaging apparatus in which the shimming means flows the current of the optimum value for each spatial component,
The static magnetic field inhomogeneity detection pulse sequence repeats a basic sequence of applying a slice encode gradient magnetic field in the slice direction, a phase encode gradient magnetic field in the phase encode direction, and a gradient magnetic field that continuously inverts in the readout direction.
The signal processing means calculates a spectrum for each voxel of the subject from an echo signal representing the static magnetic field inhomogeneity acquired so that the gradient magnetic field in the readout direction has the same polarity and the same interval in the time axis direction. A magnetic resonance imaging apparatus characterized by obtaining a static magnetic field distribution for each voxel by obtaining and converting the resonance frequency of the specific peak into a static magnetic field intensity. The magnetic resonance imaging apparatus according to claim 1.
The signal processing means develops the static magnetic field distribution for each voxel using the first and second order terms of the spherical harmonic function, and determines the optimum current value for each spatial component,
2. The magnetic resonance imaging apparatus according to claim 1, wherein the shimming means has at least eight spatial components that generate a first-order term and a second-order term of a substantially spherical harmonic function, and causes the optimum current to flow for each spatial component. The magnetic resonance imaging apparatus according to claim 1 or 2,
The magnetic resonance imaging apparatus characterized in that the measurement control means sets the number of matrices in the readout direction to be larger than the remaining two directions.

Images